FlowForge is a system that lets you design and run automated workflows connecting Web2 services (APIs, webhooks, schedules) with Web3 actions (on-chain reads, swaps, lending, borrowing).

Instead of hard-coding automations, FlowForge uses workflows: reusable, visual sequences of steps that the system executes for you.

Workflows are modelled as a directed graph of steps (called nodes) connected by edges that define execution order and data flow. This makes automations easier to reason about, replay, debug, and evolve over time.

FlowForge is designed for teams who need reliable and composable automation across Web2 and Web3:

• Crypto / DeFi teams automating trading, rebalancing, lending/borrowing, or treasury operations
• Fintech and data teams building price-driven logic using on-chain data
• Ops and engineering teams looking for a single orchestration layer instead of custom scripts and cron jobs

You don’t need to be a low-level blockchain engineer, but familiarity with DeFi concepts (swaps, wallets, gas, lending) is recommended.

A complete automation, such as:

• “Every hour, swap USDC to WETH”
• “If ETH price > X, move funds into lending”

A single step in a workflow. Common node types include:

• Trigger
  • Schedule (cron-like)
  • Manual (on demand)
  • Webhook / API event
• Swap
  • Trade tokens via aggregators (Uniswap, Relay, 1inch)
• Lending / Borrowing
  • Interact with protocols like Aave or Compound
• Price Oracle
  • Fetch prices from Chainlink or Pyth
• Logic / Conditions
  • IF / THEN branching based on data
• Web2 / API
  • Call external APIs, webhooks, or send notifications

Defines the execution order and data flow between nodes.

The on-chain address that:

• Holds funds
• Signs and submits transactions

Supported blockchain networks where workflows run, such as:

• Arbitrum One Mainnet
• Arbitrum Sepolia Testnet
• Ethereum Sepolia Testnet

• Periodic swaps (DCA strategies, treasury rebalancing)
• Automated collateral management
• Chaining oracle checks with on-chain actions

• Fetch prices from trusted oracles (Chainlink, Pyth)
• Trigger actions based on market conditions
• Standardise how price data is used across workflows

• Start workflows via schedules or webhooks
• Combine off-chain events with on-chain actions
• Replace scattered scripts and cron jobs with a single system

You specify:

• A workflow name (e.g., “Hourly USDC → WETH DCA”)
• A set of nodes and their connections
• Configuration for each node (tokens, amounts, oracle, conditions)
• How the workflow is triggered

The system checks:

• All nodes are connected in a valid execution order
• No infinite loops exist
• Required configuration (wallets, token addresses, networks) is present

When triggered:

• A scheduler or external event starts the workflow
• The engine executes nodes sequentially:
  • Fetching data (e.g., prices)
  • Evaluating conditions
  • Sending on-chain transactions when rules are met

• Each run produces a full execution trace
• Inputs and outputs of every node are recorded
• Failed steps can be inspected, debugged, or replayed

• Token swaps via Uniswap, Relay, and 1inch
• Lending & borrowing through Aave V3 and Compound V3
• Support for Arbitrum mainnet and test networks

• Chainlink and Pyth price feeds
• Use oracle data directly in workflow conditions

• Run workflows:
  • On a fixed schedule
  • On demand via UI or API
  • In response to external webhooks or events

• Build IF / THEN logic into workflows:
  • IF price > threshold → execute swap
  • IF balance exceeds X → lend excess
  • IF oracle call fails → retry or notify

• Add new node types without redesigning workflows
• Clone and adapt existing workflows for new strategies
• Future support for advanced nodes (AI/LLM, richer Web2 integrations, flow control)

• Automatic portfolio rebalancing
• Downside protection via price-based rules
• Lending idle assets to earn yield

• Rule-based strategies driven by oracle data
• Chained swaps with conditional execution

• Periodic checks on prices or positions
• Trigger notifications, hedging trades, or rebalancing

• Rapid experimentation on testnets
• Iterate on logic without writing low-level blockchain code

Setup and reference for each part of the stack:

• Backend — Project structure, setup, API, env vars, Docker
• Frontend — Project structure, setup, block system, env vars
• LLM Service — Project structure, setup, API, config, error codes

FlowForge is under active development. Features and integrations will continue to evolve.